MANOCCA: a robust and computationally efficient test of covariance in high-dimension multivariate omics data

Area of Science:

• Bioinformatics
• Statistical Genetics
• Systems Biology

Background:

• High-throughput molecular data analysis relies heavily on multivariate methods.
• Key features like covariance structure are often under-explored.
• Existing methods may miss crucial biological signals present in covariance patterns.

Purpose of the Study:

• To introduce MANOCCA (Multivariate Analysis of Conditional Covariance), a novel statistical method.
• To test the effect of predictors on the covariance matrix of multivariate outcomes.
• To capture effects missed by mean- and variance-based analyses.

Main Methods:

• MANOCCA test development and theoretical validation.
• Comparison with existing correlation-based methods using simulations.
• Dimensionality reduction via principal component analysis (PCA) for omics data.
• Application to a large cohort (Milieu Interieur) with diverse phenotypes.

Main Results:

• MANOCCA demonstrates superior calibration in omics data simulations compared to correlation-based methods.
• Optimal data dimensionality for analysis can be achieved with a limited number of principal components.
• Significant associations were found between health, lifestyle, and genetic factors and the covariance of blood and immune cell phenotypes.

Conclusions:

• MANOCCA is a powerful and well-calibrated method for analyzing covariance in high-throughput molecular data.
• The method effectively identifies biological factors influencing complex omics data relationships.
• MANOCCA offers a valuable new approach for uncovering hidden patterns in biological systems.